9.28
This commit is contained in:
parent
2c036d8504
commit
6637086547
@ -107,6 +107,6 @@ def extract_tables_main(path, output_folder):
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
path = 'C:\\Users\\Administrator\\Desktop\\招标文件\\招标test文件夹\\tmp\\ztbfile_tobidders_notice_table.docx'
|
||||
output_folder = "C:\\Users\\Administrator\\Desktop\\招标文件\\招标test文件夹\\tmp" # 前附表json文件
|
||||
path = 'C:\\Users\\Administrator\\Desktop\\货物标\\output4\\招招招标文件（一中多媒体报告厅教学设备）_tobidders_notice_part1.docx'
|
||||
output_folder = "C:\\Users\\Administrator\\Desktop\\货物标\\output4\\tmp" # 前附表json文件
|
||||
extract_tables_main(path, output_folder)
|
||||
|
||||
@ -1,210 +0,0 @@
|
||||
import json
|
||||
import docx
|
||||
import re
|
||||
import os
|
||||
from PyPDF2 import PdfReader
|
||||
from flask_app.main.截取pdf import clean_page_content,extract_common_header
|
||||
|
||||
def extract_text_from_docx(file_path):
|
||||
doc = docx.Document(file_path)
|
||||
return '\n'.join([para.text for para in doc.paragraphs])
|
||||
|
||||
|
||||
def extract_text_from_pdf(file_path):
|
||||
# 从PDF文件中提取文本
|
||||
common_header = extract_common_header(file_path)
|
||||
pdf_document = PdfReader(file_path)
|
||||
text = ""
|
||||
# 遍历每一页
|
||||
for page in pdf_document.pages:
|
||||
# 提取当前页面的文本
|
||||
page_text = page.extract_text() if page.extract_text() else ""
|
||||
# 清洗页面文本
|
||||
page_text = clean_page_content(page_text, common_header)
|
||||
# 将清洗后的文本添加到总文本中
|
||||
text += page_text+"\n"
|
||||
return text
|
||||
|
||||
def extract_section(text, start_pattern, end_phrases):
|
||||
# 查找开始模式
|
||||
start_match = re.search(start_pattern, text)
|
||||
if not start_match:
|
||||
return "" # 如果没有找到匹配的开始模式，返回空字符串
|
||||
start_index = start_match.end() # 从匹配的结束位置开始
|
||||
|
||||
# 初始化结束索引为文本总长度
|
||||
end_index = len(text)
|
||||
|
||||
# 遍历所有结束短语，查找第一个出现的结束短语
|
||||
for phrase in end_phrases:
|
||||
match = re.search(phrase, text[start_index:], flags=re.MULTILINE)
|
||||
if match:
|
||||
end_index = start_index + match.start() # 更新结束索引为匹配到的开始位置
|
||||
break # 找到第一个匹配后立即停止搜索
|
||||
|
||||
# 提取并返回从开始模式后到结束模式前的内容
|
||||
return text[start_index:end_index]
|
||||
|
||||
def compare_headings(current, new):
|
||||
# 使用过滤来确保只处理非空且为数字的部分
|
||||
current_nums = [int(num) for num in current.split('.') if num.isdigit()]
|
||||
new_nums = [int(num) for num in new.split('.') if num.isdigit()]
|
||||
|
||||
# 比较数字序列以确定标题的层次关系
|
||||
for c, n in zip(current_nums, new_nums):
|
||||
if n > c:
|
||||
return True
|
||||
elif n < c:
|
||||
return False
|
||||
|
||||
# 如果新标题有更多层次，认为是新的子章节
|
||||
return len(new_nums) > len(current_nums)
|
||||
|
||||
|
||||
def should_add_newline(content, keywords, max_length=20):
|
||||
content_str = ''.join(content).strip()
|
||||
return any(keyword in content_str for keyword in keywords) or len(content_str) <= max_length
|
||||
|
||||
def handle_content_append(current_content, line_content, append_newline, keywords):
|
||||
if append_newline:
|
||||
if should_add_newline(current_content, keywords):
|
||||
current_content.append('\n') # 添加换行符
|
||||
append_newline = False
|
||||
current_content.append(line_content)
|
||||
return append_newline
|
||||
|
||||
#对二级标题如x.x进行额外处理：如果当前处理内容包含keywords中的内容，则必须保留换行符/如果当前内容字数大于20，不保留换行。
|
||||
def parse_text_by_heading(text):
|
||||
keywords = ['包含', '以下']
|
||||
data = {}
|
||||
current_key = None
|
||||
current_content = []
|
||||
append_newline = False
|
||||
|
||||
lines = text.split('\n')
|
||||
for i, line in enumerate(lines):
|
||||
line_stripped = line.strip()
|
||||
# 匹配形如 '1.1'、'2.2.3' 等至少包含一个点的标题，并确保其前后没有字母或括号
|
||||
match = re.match(r'^(?<![a-zA-Z（(])(\d+(?:\.\d+)+)\s*(.*)', line_stripped)
|
||||
if not match:
|
||||
match = re.match(r'^(\d+\.)\s*(.+)$', line_stripped)
|
||||
|
||||
if match:
|
||||
new_key, line_content = match.groups()
|
||||
line_content = line_content.lstrip('.')
|
||||
# 检查是否应该更新当前键和内容
|
||||
if current_key is None or (compare_headings(current_key, new_key) and (
|
||||
len(current_content) == 0 or current_content[-1][-1] != '第')):
|
||||
if current_key is not None:
|
||||
# 将之前的内容保存到data中，保留第一个换行符，后续的换行符转为空字符
|
||||
content_string = ''.join(current_content).strip()
|
||||
data[current_key] = content_string.replace(' ', '')
|
||||
current_key = new_key
|
||||
current_content = [line_content]
|
||||
# 只有当标题级别为两级（如 1.1）时，才设置 append_newline 为 True
|
||||
append_newline = len(new_key.split('.')) == 2
|
||||
else:
|
||||
append_newline = handle_content_append(current_content, line_content, append_newline, keywords)
|
||||
else:
|
||||
if line_stripped:
|
||||
append_newline = handle_content_append(current_content, line_stripped, append_newline, keywords)
|
||||
|
||||
if current_key is not None:
|
||||
# 保存最后一部分内容
|
||||
content_string = ''.join(current_content).strip()
|
||||
data[current_key] = content_string.replace(' ', '')
|
||||
|
||||
return data
|
||||
|
||||
def convert_to_json(file_path, start_word, end_phrases):
|
||||
if file_path.endswith('.docx'):
|
||||
text = extract_text_from_docx(file_path)
|
||||
elif file_path.endswith('.pdf'):
|
||||
text = extract_text_from_pdf(file_path)
|
||||
else:
|
||||
raise ValueError("Unsupported file format")
|
||||
# 提取从 start_word 开始到 end_phrases 结束的内容
|
||||
text = extract_section(text, start_word, end_phrases)
|
||||
# print(text)
|
||||
parsed_data = parse_text_by_heading(text)
|
||||
return parsed_data
|
||||
|
||||
def convert_clause_to_json(input_path,output_folder,type=1):
|
||||
if not os.path.exists(input_path):
|
||||
print(f"The specified file does not exist: {input_path}")
|
||||
return ""
|
||||
if type==1:
|
||||
start_word = "投标人须知正文"
|
||||
end_phrases = [
|
||||
r'^第[一二三四五六七八九十]+章\s*评标办法', r'^评标办法前附表', r'^附录：', r'^附录一：', r'^附件：', r'^附件一：',
|
||||
r'^附表：', r'^附表一：', r'^附录:', r'^附录一:', r'^附件:', r'^附件一:', r'^附表:', r'^附表一:',
|
||||
]
|
||||
else:
|
||||
start_word = r'第[一二三四五六七八九十]+章\s*招标公告|第一卷|招标编号:|招标编号：'
|
||||
end_phrases=[r'第[一二三四五六七八九十]+章\s*投标人须知',r'投标人须知前附表']
|
||||
result = convert_to_json(input_path, start_word, end_phrases)
|
||||
file_name = "clause1.json" if type == 1 else "clause2.json"
|
||||
output_path = os.path.join(output_folder, file_name)
|
||||
with open(output_path, 'w', encoding='utf-8') as f:
|
||||
json.dump(result, f, indent=4, ensure_ascii=False)
|
||||
post_process_json(output_path)
|
||||
print(f"投标人须知正文条款提取成json文件: The data has been processed and saved to '{output_path}'.")
|
||||
return output_path
|
||||
|
||||
def post_process_json(json_file_path): #处理一级标题如'5.1'过长的内容 zbtest20
|
||||
# 读取 JSON 文件
|
||||
with open(json_file_path, 'r', encoding='utf-8') as file:
|
||||
data = json.load(file)
|
||||
|
||||
processed_data = {}
|
||||
|
||||
for key, value in data.items():
|
||||
# 检查是否是一级标题（如 '5.'），并且其值包含 '\n'
|
||||
if re.match(r'^\d+\.\s*$', key) and '\n' in value:
|
||||
# 分割标题和正文
|
||||
title, content = value.split('\n', 1)
|
||||
|
||||
# 添加原来的标题作为 '5.0'，其值为原来标题的内容（即 title）
|
||||
processed_data[key] = title.strip()
|
||||
sub_key = f"{key.rstrip('.')}." + "0" # 自动生成 '5.0'，与 '5.' 一致，保证点号的存在
|
||||
|
||||
processed_data[sub_key] = title.strip()
|
||||
|
||||
# 初始化计数器
|
||||
sub_count = 1
|
||||
|
||||
# 根据子序号 '1.' 或 '1、' 进行分割
|
||||
sub_sections = re.split(r'(\d+[\.\、])\s*', content)
|
||||
|
||||
current_sub_content = ""
|
||||
for i in range(1, len(sub_sections), 2):
|
||||
sub_number = sub_sections[i].strip() # 获取子序号
|
||||
sub_content = sub_sections[i + 1].strip() # 获取内容
|
||||
|
||||
# 生成三级标题，如 '5.0.1', '5.0.2'
|
||||
sub_key_with_number = f"{sub_key}.{sub_count}"
|
||||
processed_data[sub_key_with_number] = sub_content
|
||||
sub_count += 1
|
||||
|
||||
else:
|
||||
# 如果没有分割需求，保留原数据
|
||||
processed_data[key] = value
|
||||
|
||||
# 将修改后的数据重新写入到原来的 JSON 文件中
|
||||
with open(json_file_path, 'w', encoding='utf-8') as file:
|
||||
json.dump(processed_data, file, ensure_ascii=False, indent=4)
|
||||
|
||||
if __name__ == "__main__":
|
||||
# file_path = 'D:\\flask_project\\flask_app\\static\\output\\cfd4959d-5ea9-4112-8b50-9e543803f029\\ztbfile_tobidders_notice.pdf'
|
||||
file_path='C:\\Users\\Administrator\\Desktop\\货物标\\output3\\2-招标文件（广水市教育局封闭管理）_qualification1.pdf'
|
||||
# start_word = "投标人须知正文"
|
||||
# end_phrases = [
|
||||
# r'^第[一二三四五六七八九十]+章\s+评标办法', r'^评标办法前附表', r'^附录：', r'^附录一：', r'^附件：', r'^附件一：',
|
||||
# r'^附表：', r'^附表一：', r'^附录:', r'^附录一:', r'^附件:', r'^附件一:', r'^附表:', r'^附表一:',
|
||||
# ]
|
||||
output_folder = 'C:\\Users\\Administrator\\Desktop\\货物标\\output3\\tmp'
|
||||
try:
|
||||
output_path = convert_clause_to_json(file_path,output_folder)
|
||||
print(f"Final JSON result saved to: {output_path}")
|
||||
except ValueError as e:
|
||||
print("Error:", e)
|
||||
@ -143,7 +143,7 @@ def extract_from_notice(clause_path, type):
|
||||
return transformed_data
|
||||
|
||||
|
||||
# 假设原始数据文件路径
|
||||
#TODO:考虑5.1这种二级标题后面换行符，但是仍然存在5.1.1情况时该怎么办 再审视一下zbtest20的处理是否合理
|
||||
if __name__ == "__main__":
|
||||
file_path = 'C:\\Users\\Administrator\\Desktop\\招标文件\\招标test文件夹\\tmp\\clause1.json'
|
||||
# file_path='C:\\Users\\Administrator\\Desktop\\招标文件\\招标test文件夹\\clause1.json'
|
||||
|
||||
@ -10,21 +10,53 @@ def extract_text_from_docx(file_path):
|
||||
return '\n'.join([para.text for para in doc.paragraphs])
|
||||
|
||||
|
||||
def extract_text_from_pdf(file_path):
|
||||
# def extract_text_from_pdf(file_path):
|
||||
# # 从PDF文件中提取文本
|
||||
# common_header = extract_common_header(file_path)
|
||||
# pdf_document = PdfReader(file_path)
|
||||
# text = ""
|
||||
# # 遍历每一页
|
||||
# for page in pdf_document.pages:
|
||||
# # 提取当前页面的文本
|
||||
# page_text = page.extract_text() if page.extract_text() else ""
|
||||
# # 清洗页面文本
|
||||
# page_text = clean_page_content(page_text, common_header)
|
||||
# # 将清洗后的文本添加到总文本中
|
||||
# text += page_text+"\n"
|
||||
# return text
|
||||
def extract_text_from_pdf(file_path, start_word, end_pattern):
|
||||
# 从PDF文件中提取文本
|
||||
common_header = extract_common_header(file_path)
|
||||
pdf_document = PdfReader(file_path)
|
||||
text = ""
|
||||
# 遍历每一页
|
||||
for page in pdf_document.pages:
|
||||
# 提取当前页面的文本
|
||||
all_pages_text = []
|
||||
start_index = None
|
||||
# 处理所有页面
|
||||
for i, page in enumerate(pdf_document.pages):
|
||||
page_text = page.extract_text() if page.extract_text() else ""
|
||||
# 清洗页面文本
|
||||
page_text = clean_page_content(page_text, common_header)
|
||||
# 将清洗后的文本添加到总文本中
|
||||
text += page_text+"\n"
|
||||
return text
|
||||
cleaned_text = clean_page_content(page_text, common_header)
|
||||
|
||||
# 在第一页查找开始位置
|
||||
if i == 0 and start_index is None:
|
||||
start_match = re.search(start_word, cleaned_text, re.MULTILINE)
|
||||
if start_match:
|
||||
start_index = start_match.start()
|
||||
cleaned_text = cleaned_text[start_index:]
|
||||
|
||||
# 在最后一页查找结束位置
|
||||
if i == len(pdf_document.pages) - 1:
|
||||
for pattern in end_pattern:
|
||||
matches = list(re.finditer(pattern, cleaned_text, re.MULTILINE))
|
||||
if matches:
|
||||
end_index = matches[-1].start()
|
||||
cleaned_text = cleaned_text[:end_index]
|
||||
break
|
||||
|
||||
all_pages_text.append(cleaned_text)
|
||||
|
||||
# 合并所有页面的文本
|
||||
full_text = "\n".join(all_pages_text)
|
||||
# print(full_text)
|
||||
return full_text
|
||||
def extract_section(text, start_pattern, end_phrases):
|
||||
# 查找开始模式
|
||||
start_match = re.search(start_pattern, text)
|
||||
@ -80,9 +112,8 @@ def parse_text_by_heading(text):
|
||||
current_key = None
|
||||
current_content = []
|
||||
append_newline = False
|
||||
|
||||
lines = text.split('\n')
|
||||
for i, line in enumerate(lines):
|
||||
for i, line in enumerate(lines): #由于本身就是按行读取处理，因此保存的时候不带'\n'
|
||||
line_stripped = line.strip()
|
||||
# 匹配形如 '1.1'、'2.2.3' 等至少包含一个点的标题，并确保其前后没有字母或括号
|
||||
match = re.match(r'^(?<![a-zA-Z（(])(\d+(?:\.\d+)+)\s*(.*)', line_stripped)
|
||||
@ -94,11 +125,12 @@ def parse_text_by_heading(text):
|
||||
line_content = line_content.lstrip('.')
|
||||
# 检查是否应该更新当前键和内容
|
||||
if current_key is None or (compare_headings(current_key, new_key) and (
|
||||
len(current_content) == 0 or current_content[-1][-1] != '第')):
|
||||
len(current_content) == 0 or current_content[-1][-1] != '第')): #current_content： 内容占两行则是： ['本招标项目已具备招标条件，现对本项目施工监理进行招标，项目概况见本', '章附件一。']
|
||||
if current_key is not None:
|
||||
# 将之前的内容保存到data中，保留第一个换行符，后续的换行符转为空字符
|
||||
content_string = ''.join(current_content).strip()
|
||||
data[current_key] = content_string.replace(' ', '')
|
||||
|
||||
current_key = new_key
|
||||
current_content = [line_content]
|
||||
# 只有当标题级别为两级（如 1.1）时，才设置 append_newline 为 True
|
||||
@ -120,11 +152,11 @@ def convert_to_json(file_path, start_word, end_phrases):
|
||||
if file_path.endswith('.docx'):
|
||||
text = extract_text_from_docx(file_path)
|
||||
elif file_path.endswith('.pdf'):
|
||||
text = extract_text_from_pdf(file_path)
|
||||
text = extract_text_from_pdf(file_path,start_word,end_phrases)
|
||||
else:
|
||||
raise ValueError("Unsupported file format")
|
||||
# 提取从 start_word 开始到 end_phrases 结束的内容
|
||||
text = extract_section(text, start_word, end_phrases)
|
||||
# text = extract_section(text, start_word, end_phrases)
|
||||
# print(text)
|
||||
parsed_data = parse_text_by_heading(text)
|
||||
return parsed_data
|
||||
@ -136,13 +168,18 @@ def convert_clause_to_json(input_path,output_folder,type=1):
|
||||
if type==1:
|
||||
start_word = "投标人须知正文"
|
||||
end_phrases = [
|
||||
r'^第[一二三四五六七八九十]+章\s*评标办法', r'^评标办法前附表', r'^附录：', r'^附录一：', r'^附件：', r'^附件一：',
|
||||
r'^附表：', r'^附表一：', r'^附录:', r'^附录一:', r'^附件:', r'^附件一:', r'^附表:', r'^附表一:',
|
||||
r'^第[一二三四五六七八九十]+章\s*评标办法',
|
||||
r'^评标办法前附表',
|
||||
r'^附(?:录|件|表)(?:一)?[:：]'
|
||||
]
|
||||
else:
|
||||
start_word = r'第[一二三四五六七八九十]+章\s*招标公告|第一卷|招标编号:|招标编号：'
|
||||
end_phrases=[r'第[一二三四五六七八九十]+章\s*投标人须知',r'投标人须知前附表']
|
||||
result = convert_to_json(input_path, start_word, end_phrases)
|
||||
# 检查输出文件夹是否存在，如果不存在则创建
|
||||
if not os.path.exists(output_folder):
|
||||
os.makedirs(output_folder)
|
||||
print(f"Created output folder: {output_folder}")
|
||||
file_name = "clause1.json" if type == 1 else "clause2.json"
|
||||
output_path = os.path.join(output_folder, file_name)
|
||||
with open(output_path, 'w', encoding='utf-8') as f:
|
||||
@ -155,9 +192,7 @@ def post_process_json(json_file_path): #处理一级标题如'5.1'过长的内
|
||||
# 读取 JSON 文件
|
||||
with open(json_file_path, 'r', encoding='utf-8') as file:
|
||||
data = json.load(file)
|
||||
|
||||
processed_data = {}
|
||||
|
||||
for key, value in data.items():
|
||||
# 检查是否是一级标题（如 '5.'），并且其值包含 '\n'
|
||||
if re.match(r'^\d+\.\s*$', key) and '\n' in value:
|
||||
@ -195,14 +230,14 @@ def post_process_json(json_file_path): #处理一级标题如'5.1'过长的内
|
||||
json.dump(processed_data, file, ensure_ascii=False, indent=4)
|
||||
|
||||
if __name__ == "__main__":
|
||||
# file_path = 'D:\\flask_project\\flask_app\\static\\output\\cfd4959d-5ea9-4112-8b50-9e543803f029\\ztbfile_tobidders_notice.pdf'
|
||||
file_path='C:\\Users\\Administrator\\Desktop\\货物标\\output3\\2-招标文件（广水市教育局封闭管理）_qualification1.pdf'
|
||||
file_path = 'C:\\Users\\Administrator\\Desktop\\招标文件\\招标test文件夹\\zbtest20_tobidders_notice.pdf'
|
||||
# file_path='C:\\Users\\Administrator\\Desktop\\货物标\\output4\\招招招标文件（一中多媒体报告厅教学设备）_tobidders_notice_part1.pdf'
|
||||
# start_word = "投标人须知正文"
|
||||
# end_phrases = [
|
||||
# r'^第[一二三四五六七八九十]+章\s+评标办法', r'^评标办法前附表', r'^附录：', r'^附录一：', r'^附件：', r'^附件一：',
|
||||
# r'^附表：', r'^附表一：', r'^附录:', r'^附录一:', r'^附件:', r'^附件一:', r'^附表:', r'^附表一:',
|
||||
# ]
|
||||
output_folder = 'C:\\Users\\Administrator\\Desktop\\货物标\\output3\\tmp'
|
||||
output_folder = 'C:\\Users\\Administrator\\Desktop\\招标文件\\招标test文件夹\\tmp'
|
||||
try:
|
||||
output_path = convert_clause_to_json(file_path,output_folder)
|
||||
print(f"Final JSON result saved to: {output_path}")
|
||||
|
||||
@ -72,6 +72,6 @@ def extract_text_by_page(file_path):
|
||||
print(f"Page {page_num + 1} is empty or text could not be extracted.")
|
||||
return result
|
||||
if __name__ == '__main__':
|
||||
file_path = "C:\\Users\\Administrator\\Desktop\\货物标\\zbfiles\\招标文件正文(1)(1).pdf"
|
||||
file_path = "C:\\Users\\Administrator\\Desktop\\货物标\\output4\\招招招标文件（一中多媒体报告厅教学设备）_tobidders_notice_part2.pdf"
|
||||
res=extract_text_by_page(file_path)
|
||||
# print(res)
|
||||
@ -1,72 +1,14 @@
|
||||
import json
|
||||
import re
|
||||
#这个字典可能有嵌套，你需要遍历里面的键名，对键名作判断，而不是键值，具体是这样的：如果处于同一层级的键的数量>1并且键名全由数字或点号组成。那么就将这些序号键名全部删除，重新组织成一个字典格式的数据，你可以考虑用字符串列表来保持部分平级的数据
|
||||
#对于同级的键，如果数量>1且键名都统一，那么将键名去掉，用列表保持它们的键值
|
||||
#对于同一个字典中，可能存在若干键值对，若它们的键值都是""或者"/" 你就将它们的键值删去，它们的键名用字符串列表保存
|
||||
def test_append_newline():
|
||||
def check_append_newline(key):
|
||||
append_newline = len(key.split('.')) == 2
|
||||
return append_newline
|
||||
|
||||
#zbtest20也有问题
|
||||
def contains_number_or_index(key, value):
|
||||
# 判断值是否是数字或数字字符串
|
||||
is_number = isinstance(value, (int, float)) or (isinstance(value, str) and value.isdigit())
|
||||
# 判断键是否包含 "序号"
|
||||
contains_index = '序号' in key
|
||||
# 判断值中是否包含数字
|
||||
contains_digit = isinstance(value, str) and re.search(r'\d+', value)
|
||||
# 判断值中是否包含中文字符
|
||||
contains_chinese = isinstance(value, str) and re.search(r'[\u4e00-\u9fff]', value)
|
||||
# 如果值中包含数字但也有中文字符，则保留（返回 False）
|
||||
if contains_digit and contains_chinese:
|
||||
return False
|
||||
# 如果值是数字或包含数字，且不包含中文字符，或者键包含 "序号"，返回 True
|
||||
return is_number or contains_index or contains_digit
|
||||
# 测试用例
|
||||
test_cases = ["1.1", "1."]
|
||||
|
||||
#对于同一个字典中，可能存在若干键值对，若它们的键值都是""或者"/" 你就将它们的键值删去，它们的键名用字符串列表保存
|
||||
#如果键名是"序号"或者键值中全是数字，删去序号
|
||||
def preprocess_dict(data):
|
||||
if isinstance(data, dict):
|
||||
if len(data) > 1:
|
||||
# 检查是否所有值都是 "" 或 "/"
|
||||
if all(v == "" or v == "/" for v in data.values()):
|
||||
return list(data.keys())
|
||||
else:
|
||||
processed = {}
|
||||
for k, v in data.items():
|
||||
if not contains_number_or_index(k, v):
|
||||
processed_v = preprocess_dict(v)
|
||||
if processed_v != "": # 只添加非空值
|
||||
processed[k] = processed_v
|
||||
return processed
|
||||
else:
|
||||
return {k: preprocess_dict(v) for k, v in data.items()}
|
||||
elif isinstance(data, list):
|
||||
return [preprocess_dict(item) for item in data]
|
||||
else:
|
||||
return data
|
||||
for case in test_cases:
|
||||
result = check_append_newline(case)
|
||||
print(f"序号 '{case}': append_newline = {result}")
|
||||
|
||||
|
||||
# 测试代码
|
||||
#TODO:同一层全部都是数字才成功删除,没需求了
|
||||
input_data = {
|
||||
"符合性审查": {
|
||||
"说明": "1ha",
|
||||
"www":"哈哈",
|
||||
"审查标准": [
|
||||
{
|
||||
"序号": 1,
|
||||
"内容": "投标总报价超过项目（分包）预算金额或最高限价的；"
|
||||
},
|
||||
{
|
||||
"序号": 2,
|
||||
"内容": "《投标书》、《法定代表人授权书》、《开标一览表（含明细）》未提供或不符合招标文件要求的；"
|
||||
},
|
||||
{
|
||||
"序号": 3,
|
||||
"内容": "工期（服务期限）、质保期不符合招标文件要求的；"
|
||||
},
|
||||
]
|
||||
}
|
||||
}
|
||||
pred=preprocess_dict(input_data)
|
||||
print(json.dumps(pred, ensure_ascii=False, indent=4))
|
||||
# processed_data = process_dict(pred)
|
||||
# print(json.dumps(processed_data, ensure_ascii=False, indent=4))
|
||||
# 运行测试
|
||||
test_append_newline()
|
||||
184
flask_app/货物标/投标人须知正文条款提取成json文件货物标版.py
Normal file
184
flask_app/货物标/投标人须知正文条款提取成json文件货物标版.py
Normal file
@ -0,0 +1,184 @@
|
||||
import json
|
||||
import docx
|
||||
import re
|
||||
import os
|
||||
from PyPDF2 import PdfReader
|
||||
from flask_app.main.截取pdf import clean_page_content,extract_common_header
|
||||
|
||||
def extract_text_from_docx(file_path):
|
||||
doc = docx.Document(file_path)
|
||||
return '\n'.join([para.text for para in doc.paragraphs])
|
||||
|
||||
|
||||
def extract_text_from_pdf(file_path, start_word, end_pattern):
|
||||
# 从PDF文件中提取文本
|
||||
common_header = extract_common_header(file_path)
|
||||
pdf_document = PdfReader(file_path)
|
||||
all_pages_text = []
|
||||
start_index = None
|
||||
# 处理所有页面
|
||||
for i, page in enumerate(pdf_document.pages):
|
||||
page_text = page.extract_text() if page.extract_text() else ""
|
||||
cleaned_text = clean_page_content(page_text, common_header)
|
||||
|
||||
# 在第一页查找开始位置
|
||||
if i == 0 and start_index is None:
|
||||
start_match = re.search(start_word, cleaned_text, re.MULTILINE)
|
||||
if start_match:
|
||||
start_index = start_match.start()
|
||||
cleaned_text = cleaned_text[start_index:]
|
||||
|
||||
# 在最后一页查找结束位置
|
||||
if i == len(pdf_document.pages) - 1:
|
||||
for pattern in end_pattern:
|
||||
matches = list(re.finditer(pattern, cleaned_text, re.MULTILINE))
|
||||
if matches:
|
||||
end_index = matches[-1].start()
|
||||
cleaned_text = cleaned_text[:end_index]
|
||||
break
|
||||
|
||||
all_pages_text.append(cleaned_text)
|
||||
|
||||
# 合并所有页面的文本
|
||||
full_text = "\n".join(all_pages_text)
|
||||
# print(full_text)
|
||||
return full_text
|
||||
|
||||
def compare_headings(current, new):
|
||||
# 使用过滤来确保只处理非空且为数字的部分
|
||||
current_nums = [int(num) for num in current.split('.') if num.isdigit()]
|
||||
new_nums = [int(num) for num in new.split('.') if num.isdigit()]
|
||||
|
||||
# 比较数字序列以确定标题的层次关系
|
||||
for c, n in zip(current_nums, new_nums):
|
||||
if n > c:
|
||||
return True
|
||||
elif n < c:
|
||||
return False
|
||||
|
||||
# 如果新标题有更多层次，认为是新的子章节
|
||||
return len(new_nums) > len(current_nums)
|
||||
|
||||
|
||||
def should_add_newline(content, keywords, max_length=20):
|
||||
content_str = ''.join(content).strip()
|
||||
return any(keyword in content_str for keyword in keywords) or len(content_str) <= max_length
|
||||
|
||||
def handle_content_append(current_content, line_content, append_newline, keywords):
|
||||
if append_newline:
|
||||
if should_add_newline(current_content, keywords):
|
||||
current_content.append('\n') # 添加换行符
|
||||
append_newline = False
|
||||
current_content.append(line_content)
|
||||
return append_newline
|
||||
|
||||
"""
|
||||
保存换行符的具体逻辑：
|
||||
|
||||
对于二级标题（如 1.1），如果其后的内容包含关键词或内容较短（<=20字符），会在内容前添加一个换行符。
|
||||
这个换行符会被保留在 current_content 列表中。
|
||||
当处理下一个标题时，之前的内容（包括可能存在的换行符）会被合并并保存到 data 字典中。
|
||||
"""
|
||||
#提取json主函数
|
||||
def parse_text_by_heading(text):
|
||||
keywords = ['包含', '以下']
|
||||
data = {}
|
||||
current_key = None
|
||||
current_content = []
|
||||
append_newline = False
|
||||
|
||||
lines = text.split('\n') #['一、说明', '1.适用范围', '招标文件仅适用于第一章“投标邀请书”中所述项目的货物、工程及服务的采购。']
|
||||
for i, line in enumerate(lines): #由于本身就是按行读取处理，因此保存的时候不带'\n'
|
||||
line_stripped = line.strip()
|
||||
|
||||
# 匹配中文数字标题，如 "一、说明"
|
||||
chinese_match = re.match(r'^([一二三四五六七八九十]+、)\s*(.+)$', line_stripped)
|
||||
if chinese_match:
|
||||
chinese_key, chinese_value = chinese_match.groups()
|
||||
chinese_key = chinese_key.rstrip('、') # 移除顿号
|
||||
data[chinese_key] = chinese_value
|
||||
current_key = None
|
||||
current_content = []
|
||||
continue
|
||||
|
||||
# 匹配形如 '1.1'、'2.2.3' 等至少包含一个点的标题，并确保其前后没有字母或括号
|
||||
match = re.match(r'^(?<![a-zA-Z（(])(\d+(?:\.\d+)+)\s*(.*)', line_stripped)
|
||||
if not match:
|
||||
match = re.match(r'^(\d+\.)\s*(.+)$', line_stripped)
|
||||
|
||||
if match:
|
||||
new_key, line_content = match.groups()
|
||||
line_content = line_content.lstrip('.')
|
||||
# 检查是否应该更新当前键和内容
|
||||
if current_key is None or (compare_headings(current_key, new_key) and (
|
||||
len(current_content) == 0 or current_content[-1][-1] != '第')):
|
||||
if current_key is not None:
|
||||
# 将之前的内容保存到data中，保留第一个换行符，后续的换行符转为空字符
|
||||
# print(current_content)
|
||||
content_string = ''.join(current_content).strip()
|
||||
# print(content_string)
|
||||
data[current_key] = content_string.replace(' ', '')
|
||||
current_key = new_key
|
||||
current_content = [line_content]
|
||||
# 只有当标题级别为两级（如 1.1）时，才设置 append_newline 为 True #TODO:一级标题也是
|
||||
append_newline = len(new_key.split('.')) == 2
|
||||
print(new_key)
|
||||
else:
|
||||
append_newline = handle_content_append(current_content, line_content, append_newline, keywords)
|
||||
else:
|
||||
if line_stripped:
|
||||
append_newline = handle_content_append(current_content, line_stripped, append_newline, keywords)
|
||||
|
||||
if current_key is not None:
|
||||
# 保存最后一部分内容
|
||||
content_string = ''.join(current_content).strip()
|
||||
data[current_key] = content_string.replace(' ', '')
|
||||
|
||||
return data
|
||||
|
||||
# def convert_to_json(file_path, start_word, end_phrases):
|
||||
# if file_path.endswith('.docx'):
|
||||
# text = extract_text_from_docx(file_path)
|
||||
# elif file_path.endswith('.pdf'):
|
||||
# text = extract_text_from_pdf(file_path,start_word,end_phrases)
|
||||
# else:
|
||||
# raise ValueError("Unsupported file format")
|
||||
# # print(text)
|
||||
# parsed_data = parse_text_by_heading(text)
|
||||
# return parsed_data
|
||||
|
||||
def convert_clause_to_json(input_path,output_folder,type=1):
|
||||
if not os.path.exists(input_path):
|
||||
print(f"The specified file does not exist: {input_path}")
|
||||
return ""
|
||||
if type == 1:
|
||||
start_word = r'^\s*[（(]?\s*[一1]\s*[)）]?\s*[、.]*\s*(说\s*明|总\s*则)'
|
||||
end_pattern = [r'^第[一二三四五六七八九十百千]+(?:章|部分)\s*[\u4e00-\u9fff]+']
|
||||
else:
|
||||
start_word = r'第[一二三四五六七八九十]+章\s*招标公告|第一卷|招标编号:|招标编号：'
|
||||
end_pattern=[r'第[一二三四五六七八九十]+章\s*投标人须知',r'投标人须知前附表']
|
||||
text = extract_text_from_pdf(file_path, start_word, end_pattern)
|
||||
result = parse_text_by_heading(text)
|
||||
# result = convert_to_json(input_path, start_word, end_pattern)
|
||||
# 检查输出文件夹是否存在，如果不存在则创建
|
||||
if not os.path.exists(output_folder):
|
||||
os.makedirs(output_folder)
|
||||
print(f"Created output folder: {output_folder}")
|
||||
file_name = "clause1.json" if type == 1 else "clause2.json"
|
||||
output_path = os.path.join(output_folder, file_name)
|
||||
with open(output_path, 'w', encoding='utf-8') as f:
|
||||
json.dump(result, f, indent=4, ensure_ascii=False)
|
||||
print(f"投标人须知正文条款提取成json文件: The data has been processed and saved to '{output_path}'.")
|
||||
return output_path
|
||||
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
# file_path = 'D:\\flask_project\\flask_app\\static\\output\\cfd4959d-5ea9-4112-8b50-9e543803f029\\ztbfile_tobidders_notice.pdf'
|
||||
file_path='C:\\Users\\Administrator\\Desktop\\货物标\\output4\\招招招标文件（一中多媒体报告厅教学设备）_tobidders_notice_part2.pdf'
|
||||
output_folder = 'C:\\Users\\Administrator\\Desktop\\货物标\\output4\\tmp1'
|
||||
try:
|
||||
output_path = convert_clause_to_json(file_path,output_folder)
|
||||
print(f"Final JSON result saved to: {output_path}")
|
||||
except ValueError as e:
|
||||
print("Error:", e)
|
||||
@ -161,12 +161,16 @@ def process_input(input_path, output_folder, begin_pattern, begin_page, end_patt
|
||||
return generated_files
|
||||
|
||||
|
||||
def extract_pages_generic(pdf_document, begin_pattern, end_pattern, begin_page, common_header, exclusion_pattern=None):
|
||||
def extract_pages_generic(pdf_document, begin_pattern, end_pattern, begin_page, common_header, exclusion_pattern=None,output_suffix="normal"):
|
||||
start_page = None
|
||||
end_page = None
|
||||
for i, page in enumerate(pdf_document.pages):
|
||||
text = page.extract_text() or ""
|
||||
cleaned_text = clean_page_content(text, common_header)
|
||||
if output_suffix=="tobidders_notice":
|
||||
if exclusion_pattern and re.search(exclusion_pattern, cleaned_text) and start_page is not None:
|
||||
continue
|
||||
else:
|
||||
if exclusion_pattern and re.search(exclusion_pattern, cleaned_text):
|
||||
continue
|
||||
if start_page is None and re.search(begin_pattern, cleaned_text) and i > begin_page:
|
||||
@ -183,8 +187,9 @@ def extract_pages(pdf_path, output_folder, begin_pattern, begin_page, end_patter
|
||||
pdf_document = PdfReader(pdf_path)
|
||||
exclusion_pattern = None
|
||||
if output_suffix == "tobidders_notice":
|
||||
exclusion_pattern = re.compile(r'文件的构成|文件的组成|须对应|需对应|须按照|需按照|须根据|需根据')
|
||||
start_page, mid_page, end_page = extract_pages_tobidders_notice(pdf_document, begin_pattern, end_pattern,
|
||||
begin_page, common_header)
|
||||
begin_page, common_header,exclusion_pattern)
|
||||
if start_page is None or mid_page is None or end_page is None:
|
||||
print(f"first: {output_suffix} 未找到起始或结束页在文件 {pdf_path} 中！尝试备用提取策略。")
|
||||
return extract_pages_twice_tobidders_notice(pdf_path, output_folder, output_suffix, common_header)
|
||||
@ -208,13 +213,15 @@ def extract_pages(pdf_path, output_folder, begin_pattern, begin_page, end_patter
|
||||
print(f"Error processing {pdf_path}: {e}")
|
||||
return None
|
||||
|
||||
def extract_pages_tobidders_notice(pdf_document, begin_pattern, end_pattern, begin_page, common_header):
|
||||
def extract_pages_tobidders_notice(pdf_document, begin_pattern, end_pattern, begin_page, common_header,exclusion_pattern):
|
||||
start_page = None
|
||||
mid_page = None
|
||||
end_page = None
|
||||
for i, page in enumerate(pdf_document.pages):
|
||||
text = page.extract_text() or ""
|
||||
cleaned_text = clean_page_content(text, common_header)
|
||||
if exclusion_pattern and re.search(exclusion_pattern, cleaned_text) and mid_page is not None:
|
||||
continue
|
||||
if start_page is None and re.search(begin_pattern, cleaned_text) and i > begin_page:
|
||||
start_page = i
|
||||
if start_page is not None and mid_page is None and re.search(
|
||||
@ -266,6 +273,7 @@ def extract_pages_twice_tobidders_notice(pdf_path, output_folder, output_suffix,
|
||||
r'^第[一二三四五六七八九十百千]+(?:章|部分)\s*[\u4e00-\u9fff]+'
|
||||
)
|
||||
pdf_document = PdfReader(pdf_path)
|
||||
exclusion_pattern = re.compile(r'文件的构成|文件的组成|须对应|需对应|须按照|需按照|须根据|需根据')
|
||||
# 提取第一部分
|
||||
start_page1, end_page1 = extract_pages_generic(pdf_document, begin_pattern, end_pattern, -1, common_header)
|
||||
if start_page1 is None or end_page1 is None:
|
||||
@ -273,7 +281,7 @@ def extract_pages_twice_tobidders_notice(pdf_path, output_folder, output_suffix,
|
||||
return None, None
|
||||
# 提取第二部分
|
||||
start_page2 = end_page1 # 第二部分的开始页就是第一部分的结束页
|
||||
_, end_page2 = extract_pages_generic(pdf_document, end_pattern, end_pattern, start_page2 - 1, common_header)
|
||||
_, end_page2 = extract_pages_generic(pdf_document, end_pattern, end_pattern, start_page2 - 1, common_header,exclusion_pattern)
|
||||
if end_page2 is None:
|
||||
print(f"second: {output_suffix} 未找到第二部分的结束页在文件 {pdf_path} 中！")
|
||||
return None, None
|
||||
@ -390,7 +398,7 @@ def truncate_pdf_multiple(input_path, output_folder):
|
||||
|
||||
# TODO:交通智能系统和招标(1)(1)文件有问题 sele=4的时候excludsion有问题
|
||||
if __name__ == "__main__":
|
||||
input_path = "C:\\Users\\Administrator\\Desktop\\货物标\\zbfiles\\2-招标文件（2020年广水市中小学教师办公电脑系统及多媒体“班班通”设备采购安装项目）.pdf"
|
||||
input_path = "C:\\Users\\Administrator\\Desktop\\货物标\\zbfiles"
|
||||
output_folder = "C:\\Users\\Administrator\\Desktop\\货物标\\output4"
|
||||
# truncate_pdf_multiple(input_path,output_folder)
|
||||
selection = 4 # 例如：1 - 商务技术服务要求, 2 - 评标办法, 3 - 资格审查后缀有qualification1和qualification2 4.投标人须知前附表
|
||||
|
||||
Loading…
x
Reference in New Issue
Block a user